Project summary/Abstract: Frontline physicians need reliable solutions to discriminate benign from sinister causes of dizziness. Current strategies to discriminate rely on either clinical dogma or brain computerized tomography scans, but these strategies have major flaws and are linked to misdiagnosis. Among dizziness presentations, the acute vestibular syndrome - a common category of acute-onset dizziness - is the presentation where physician decisions are the most crucial. On the one hand, the most common cause of the acute vestibular syndrome is an entirely benign peripheral vestibular viral syndrome (i.e., so called vestibular neuritis or labyrinthitis), but on the other hand the cause may be a life threatening stroke. No previous study accurately estimates basic information such as the proportion of stroke etiology in acute vestibular syndrome presentations. In addition, no previous study uses rigorous methodology to estimate the risk of stroke in an individual patient. Dizziness presentations are different from other common diagnostic dilemmas (e.g., chest pain and headache presentations) because no laboratory, electrophysiological, or neuro-imaging test is a valid and practical discriminator. Though magnetic resonance imaging (MRI) is a sensitive test for stroke, it is not a practical test to use as a discriminator in the real world emergency settings. A stroke risk score - calculated from routinely collected clinical information - may be a practical discriminator. Another potential discriminator may be the 'head thrust test,' a simple bedside test of the vestibular system. However, a stroke risk score has not been developed in this unique presentation and the head thrust test has not undergone rigorous study to determine its operating characteristics in this important clinical setting. In this R18 application, we aim to take the initial steps to optimizing value and safety of health care in acute dizziness presentations. A representative sample will be recruited and all subjects will undergo a standardized clinical evaluation blinded to MRI results. Using MRI with diffusion weighted sequences as a gold standard test for stroke etiology, we plan to develop a statistical model assessing the risk of stroke in patients presenting with the acute vestibular syndrome. Model- based predictions will be based on clinical measurements and the head thrust test. The model will be used to assess the prevalence of stroke in the target patient population and the significance of the head thrust test as a risk predictor with and without clinical covariates. Using specific sensitivity and specificity tradeoff parameters, we will formulate a decision rule based on the model. Internal model validation will be performed to obtain unbiased estimates of operating characteristics. The reliability of the head thrust test will also be measured. This will help estimate model performance in future samples. This demonstration project will lay the groundwork for future multi-center validation and implementation studies that could ultimately improve the value of care and patient safety, while also fostering the appropriate use of processes of care and reducing unnecessary expenditures.